dolphinteam/DophinWhistle-Classification-Finetuning

---
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- config_name: unbalanced
data_files:
- split: train
path: unbalanced/train-*
- split: test
path: unbalanced/test-*
- config_name: all
data_files:
- split: train
path: all/train-*
- split: test
path: all/test-*
---

# DolphinWhistle-Classification-Finetuning

This dataset contains embedded dolphin whistle segments with identity labels for finetuning audio encoders.

## Available configs

- `default` (`balanced`): 2400 train / 600 test / 6 labels
- `unbalanced`: 2790 train / 698 test / 10 labels
- `all`: 6683 train / 1671 test / 10 labels

## Label sets

- `default` (`balanced`): NSW_9, SW_Luna, SW_Nana, SW_Neo, SW_Nikita, SW_Yosefa
- `unbalanced`: NSW_3, NSW_6, NSW_9, SW_Dana, SW_Luna, SW_Nana, SW_Neo, SW_Nikita, SW_Shy, SW_Yosefa
- `all`: NSW_3, NSW_6, NSW_9, SW_Dana, SW_Luna, SW_Nana, SW_Neo, SW_Nikita, SW_Shy, SW_Yosefa

## Example

```python
from datasets import load_dataset, Audio

balanced = load_dataset("dolphinteam/DophinWhistle-Classification-Finetuning")
unbalanced = load_dataset("dolphinteam/DophinWhistle-Classification-Finetuning", "unbalanced")
all_cfg = load_dataset("dolphinteam/DophinWhistle-Classification-Finetuning", "all")

unbalanced["train"] = unbalanced["train"].cast_column("audio", Audio(decode=False))
```